orange squeezer noise

[kvb]

After spending some time putting the OS together, studying the schematic, and fixing some possible mistakes in the layout provided in Dragonfly's collection I am having a problem with noise. I'm using a TL072 and mpf102 fets.

The voltages for the components match those found on in the build instuctions of GGG's OS page.

When I turn the trimmer down the volume drops but a hiss remains. With the trimmer turned up more the noise becomes impossible to miss.  While playing, the notes are clear. As the note decays the noise gradually increases.

The OS seems to be working, as one can hear the initial pluck being squashed, and then later some of the AM that is also mentioned at GGG.

When I try to ground the op amps other input, it kills the whole effect.

Any other suggestions or similar problems?              I imagine I should order a pcb.

[mdh]

This is just a shot in the dark, but if you look at the GGG layout, http://www.generalguitargadgets.com/diagrams/orangesqueezer_lo.gif, you can see that not only is the unused non-inverting input grounded, also the unused inverting input is tied to the unused output. If you haven't done that as well, it might be worth trying (I haven't looked at Dragonfly's layout, so maybe this is already taken care of).

[Mark Hammer]

Sigh....

That is EXACTLY how compressors work.

The goal of the compressor is to provide a constant output level for as long as possible.  This is accomplished by boosting the level for the quiet parts, and turning the level down for the louder parts.  The unit/circuit has no means for distinguishing between "really, really soft playing" and "I'm not playing anything right at this moment so just turn off".  As a result, when you stop playing, the gain is increased in response to this "quiet part".  Unfortunately, when you stop playing, any residual hiss from the components at the input, or coming in from the instrument itself, anything acquired over long-ish and lesser quality cables, or from pedals between the guitar and compressor, finds itself heavily amplified, and we hear that as this annoying hiss that creeps in.

How do you minimize it?

1) Using lower noise components on the input of the compressor can be helpful on occasion.  So, 1% metal film resistors anywhere that has a direct path to the input of the op-amp is a good idea (not needed on the output or sidechain).

2) Making sure that your input signal to the compressor itself is as noise-free as possible.

3) Making sure that the input to the compressor is relatively "hot", so that the compressor spends as little time as possible at maximum gain.


All of this above is generic advice about compressors.  The OS is one subclass of compressors that does not adjust gain, but rather applies constant gain (x23 in the stock unit) to an input signal that is attenuated in a manner proportional to input level.  The 82k resistor and FET form a kind of input level control.  When the input signal is hot, that results in the resistance formed by the FET going lower, AS IF one were turning down the input level.  Apply the same amount of gain to a lower input signal and the output is lower, right?  So the net effect is the same as varying the gain.

One of the perennial problems of simple inexpensive compresors is that one needs to balance off this constant issue of hiss against the frequently observed loss of "bite" and presence in compressed guitar.  Shave of the treble somewhere and you may lose hiss but you lose a LOT of presence.  Stick in a component here or there to regain the presence, and you can inadvertently end up boosting the hiss.  One solution is to shave off hiss at the input.  You can try sticking something like a 1000pf cap to ground in parallel with the input terminating resistor (the 4M7 one).

[kvb]

It's not that I am unfamiliar with compressors - just this one.  It's hard to know what to expect when I have nothing to compare it to except a different compressor.  Now I can see that the OS is meant to do things its own special way.

Things seemed different when I was able to listen to the effect through a speaker, instead of headphones.

I tried a pedal in front of the OS, and it did seem a little more consistent.
Your explanation makes sense of why Analog Man tells people to put their overdrive in front of their compressor.

[Mark Hammer]

Personally, I *wouldn't* put any high gain pedal in front of a compressor.  True, it provides a hotter signal, as I noted, but internal to an overdrive is the *risk* (and I say risk, not "certainty") of generating hiss at its output (if on), which now gets boosted by the compressor when you stop playing.  If it is an overdrive with good internal noise control, and what you are feeding it also has decent noise control (i.e., shielded nicely, low noise PUs, decent cable from the guitar), then sticking a compressor after the overdrive allows you to get whatever touch sensitivity there might be in the overdrive, but with the constant level of compression.  Certainly, if it is a more touch-sensitive overdrive, sticking a compressor in front and using moderate to heavy compression will deprive you of being able to milk that overdrive touch-sensitivity.

But here's the thing.  Most clipping devices entail some restriction of dynamic range, so what you are feeding the compressor is already compressed to some extent.  Not necessarily so compressed that the compressor ends up adding/doing nothing of its own, but compressed enough that one does not achieve any real added value by throwing compression into the mix.  Just about anything that uses a dynamics-sensing sidechain (limiter, compressor, noise-gate, envelope-controlled phaser/tremolo/filter) benefits from having more dynamics in the input signal, and some (e.g., Q-Tron +) are even designed to permit extracting the envelope before any distortion or whatever is done, with the resulting envelope voltage passed on further down the processing chain.

So, I'm not completely refuting Mike's suggestion.  I am simply qualifying it and saying that there will be some regularly-occurring circumstances where overdrive->compression is a bad or moot idea, and some occasional circumstances where it isn't a bad idea and adds its own charm or flavour to the sound.  The best general advice for using compressors, however (and I stand by this), is to always strive to provide the compressor with the most noise-free signal you possibly can.  If yu can achieve that, then you can feel free to perform any sort of sick and twisted aural experiments you wish, and your compressor will behave politely.

As for speakers vs headphones, the typical guitar amp speaker with its limited high end (most will roll off sharply above 6khz) hides a multitude of audio sins, one of which is hiss.

Incidentally, the OS is very much like many commercial compressor/limiters.  A number of "optical" compressors use their photocell exactly the same way that the OS uses the FET, and that is as the wiper-to-ground leg of a composite simulated pot (where the input-to-wiper leg is a fixed resistor).  I don't have enough experience with these things to have an opinion about whether variable gain reduction has some set of constant distinguishing characteristics that set it apart from variable input attenuation.